frogid_dataset

Study Extent

FrogID submissions have come from across Australia but, not surprisingly, are biased towards populated areas, with large areas of Australia, particularly in remote areas, lacking FrogID records. Despite this bias, the spatial coverage of this project encompasses the continent of Australia, with frog records from 7,635,905 km2 (99%) of Australia’s landmass (measured as a minimum convex polygon enclosing all occurrences, excluding ocean).

Sampling

FrogID collects data via a free smartphone app (iOS and Android). Recordings are 20-60 seconds in MPEG AAC audio (mp4a) files. The time, date and geographic location (latitude, longitude and an estimate of precision of geographic location) are automatically added by the app at the time of recording. Each recording has an estimate of precision and, depending on the question, these records may influence results. As such, for records that rely heavily on geographic precision, we recommend filtering to records which have an estimate of geographic uncertainty of < 3000 meters. After recordings are submitted, they are stored in a cloud-based Content Management System (CMS), before being validated. FrogID validators, experts in identifying frog species by their calls, then use the audio and associated information, plus a reference call library, to identify the frog species calling in the recording. One submission can have multiple frog species calling within it. After these processes, we are left with a presence-only dataset of frog species in Australia. For a more detailed overview of methodology and design aspects see Rowley et al. (2019).

Method steps

1. For a more detailed overview of methodology and design aspects see Rowley et al. (2019).

Australia